Terminal-related CLI agent task on TerminalBench 1.0
54.4AccuracyGPT-5.2
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GPT-5.2Size=-, Scaffold=Terminus 22026.06 | 54.4 | — | |
| GPT-5.2Model Size=-, Agent Scaffolds=Terminus-22026.04 | 54.38 | — | |
| Gemini-3-Flash-PreviewModel Size=-, Agent Scaffolds=Terminus-22026.04 | 53.72 | — | |
| Claude-Sonnet-4.5Model Size=-, Agent Scaffolds=Terminus-22026.04 | 51 | — | |
| GLM-4.7Size=358B, Scaffold=Terminus 22026.06 | 48.8 | — | |
| GLM-4.7Model Size=358B, Agent Scaffolds=Terminus-22026.04 | 48.75 | — | |
| Claude-Opus-4.5Model Size=-, Agent Scaffolds=Terminus-22026.04 | 47.5 | — | |
| Claude-Opus-4.5Size=-, Scaffold=Terminus 22026.06 | 47.5 | — | |
| Gemini 3 Pro PreviewSize=-, Scaffold=Terminus 22026.06 | 46.4 | — | |
| Gemini-3-Pro-PreviewModel Size=-, Agent Scaffolds=Terminus-22026.04 | 46.35 | — | |
| DeepSeek-V3.2Model Size=685B, Agent Scaffolds=TACO+Terminus-22026.04 | 46.25 | — | |
| LiberCoder-235BSize=235B, Scaffold=OpenHands2026.06 | 46.1 | — | |
| TACOBackbone=MiniMax-2.52026.04 | 45.25 | 23.43 | |
| Minimax-M2.5Model Size=230B, Agent Scaffolds=TACO+Terminus-22026.04 | 45.25 | — | |
| Kimi-K2-InstructSize=1T, Scaffold=Terminus 22026.06 | 44.6 | — | |
| Kimi-K2-InstructModel Size=1T, Agent Scaffolds=Terminus-22026.04 | 44.59 | — | |
| DeepSeek-V3.2Model Size=685B, Agent Scaffolds=Terminus-22026.04 | 43.93 | — | |
| Minimax-M2.1Model Size=229B, Agent Scaffolds=Terminus-22026.04 | 42.5 | — | |
| Minimax-M2.1Size=229B, Scaffold=Terminus 22026.06 | 42.5 | — | |
| BaselineBackbone=MiniMax-2.52026.04 | 42.3 | 29.74 | |
| Minimax-M2.5Model Size=230B, Agent Scaffolds=Terminus-22026.04 | 42.3 | — | |
| LiberCoder-32BSize=32B, Scaffold=OpenHands2026.06 | 38.9 | — | |
| CLI-Universe-32BSize=32B, Scaffold=Terminus 22026.06 | 38.8 | — | |
| Qwen3-Coder-480B-InstructModel Size=480B, Agent Scaffolds=TACO+Terminus-22026.04 | 38.5 | — | |
| Qwen3-Coder-480BModel Size=480B, Agent Scaffolds=Terminus-22026.04 | 37.5 | — | |
| Qwen3-Coder-480BSize=480B, Scaffold=Terminus 22026.06 | 37.5 | — | |
| GPT-5.1Model Size=-, Agent Scaffolds=Terminus-22026.04 | 35.5 | — | |
| TerminalTraj-32BSize=32B, Scaffold=Terminus 22026.06 | 35.3 | — | |
| SkillSynth-32BSize=32B, Scaffold=Terminus 22026.06 | 33.8 | — | |
| DeepSeek-V3.1-Nex-N1Model Size=685B, Agent Scaffolds=OpenHands2026.04 | 31.56 | — | |
| TerminalTraj-14BSize=14B, Scaffold=Terminus 22026.06 | 28.9 | — | |
| Qwen3-32B-Nex-N1Model Size=32B, Agent Scaffolds=OpenHands2026.04 | 28.75 | — | |
| CLI-Universe-14BSize=14B, Scaffold=Terminus 22026.06 | 27.5 | — | |
| Qwen3-30B-A3B-Nex-N1Model Size=30B, Agent Scaffolds=OpenHands2026.04 | 25 | — | |
| Qwen3-Coder-30B-A3B-InstructModel Size=30B, Agent Scaffolds=Terminus-22026.04 | 23.8 | — | |
| Qwen3-Coder-30B-A3BSize=30B, Scaffold=Terminus 22026.06 | 23.8 | — | |
| TerminalTraj-7BSize=7B, Scaffold=Terminus 22026.06 | 23 | — | |
| SkillSynth-14BSize=14B, Scaffold=Terminus 22026.06 | 22.9 | — | |
| CLI-Universe-8BSize=8B, Scaffold=Terminus 22026.06 | 19.1 | — | |
| SkillSynth-8BSize=8B, Scaffold=Terminus 22026.06 | 17.1 | — | |
| Qwen3-235B-A22B-InstructModel Size=235B, Agent Scaffolds=Terminus-22026.04 | 15 | — | |
| Qwen3-32B-InstructModel Size=32B, Agent Scaffolds=TACO+Terminus-22026.04 | 14.13 | — | |
| Qwen3-14BSize=14B, Scaffold=Terminus 22026.06 | 11.6 | — | |
| Qwen3-32B-InstructModel Size=32B, Agent Scaffolds=OpenHands2026.04 | 11.25 | — | |
| Qwen3-32B-InstructModel Size=32B, Agent Scaffolds=Terminus-22026.04 | 11.25 | — | |
| Qwen3-14B-InstructModel Size=14B, Agent Scaffolds=TACO+Terminus-22026.04 | 11.25 | — | |
| Qwen3-32BSize=32B, Scaffold=Terminus 22026.06 | 11.2 | — | |
| Qwen3-8B-InstructModel Size=8B, Agent Scaffolds=TACO+Terminus-22026.04 | 9.22 | — | |
| Qwen3-8B-InstructModel Size=8B, Agent Scaffolds=Terminus-22026.04 | 8.86 | — | |
| Qwen3-8BSize=8B, Scaffold=Terminus 22026.06 | 8.1 | — | |
| Qwen3-14B-InstructModel Size=14B, Agent Scaffolds=Terminus-22026.04 | 5.23 | — |